How to define and use persistent storage
Certain applications can benefit from advanced storage configurations and if a charmed operator exists for such an application Juju can declare such requirements both at deploy time and during the lifetime of the application.
The level of sophistication is limited by the charm; a charmed operator may support multiple storage options (e.g. persistent storage, additional cache). All this allows the user to allocate resources at a granular level. This page will refer to the PostgreSQL charm and the Ceph OSD charm.
The Ceph examples used here are based on the Ceph cluster described in the document Installing Ceph.
- Storage management commands
- Storage constraints
- Storage pools
- Dynamic storage
- Storage providers
- Writing charms
Outside of application deployment, Juju also has a wide array of storage management abilities.
- Creates and attaches a storage instance to a unit.
- Attaches an existing storage instance to a unit.
- Creates or defines a storage pool.
- Detaches a storage instance from a unit. Storage is preserved.
- Imports a filesystem into a model.
- Removes a storage instance from a model. Storage is destroyed.
- Shows details of a storage instance.
- Lists all storage instances in a model.
- Lists all storage pools in a model.
Several properties are used to dictate how storage is allocated:
- ‘pool’: class of storage
- ‘size’: size of each volume
- ‘count’: number of volumes
The default pool (e.g. ‘ebs’ for AWS, ‘cinder’ for OpenStack) is given by:
juju model-config storage-default-block-source
These properties are specified as constraints with the
juju deploy or
juju add-storage commands.
juju deploy <charm> [--storage <label>=<pool>,<size>,<count>]
labelis a string taken from the charmed operator itself. It encapsulates a specific storage option/feature. Sometimes called a store.
--storagemay be specified multiple times, to support multiple labels.
If at least one constraint is specified the following default values come into effect:
- ‘pool’ = the default pool (see above)
- ‘size’ = determined from the charm’s minimum storage size, or 1GiB if the charmed operator does not specify a minimum
- ‘count’ = the minimum number required by the charm, or ‘1’ if the storage is optional
In the absence of any storage constraints, the storage will be put on the root filesystem.
juju add-storage <unit> <label>[=<pool>,<size>,<count>]
juju add-unit the storage parameters used are taken from the
juju deploy command corresponding to the unit.
To deploy PostgreSQL with one instance (count) of 100GiB, via the charm’s ‘pgdata’ storage label, using the default storage pool:
juju deploy postgresql --storage pgdata=100G
Assuming an AWS model, a more explicit, but equivalent, command is:
juju deploy postgresql --storage pgdata=ebs,100G,1
juju storage-pools command to list the predefined storage pools as well as any custom ones that may have been created with the
juju create-storage-pool command:
Here is sample output for a newly-added AWS model:
Name Provider Attributes ebs ebs ebs-ssd ebs volume-type=ssd loop loop rootfs rootfs tmpfs tmpfs
The name given to a default storage pool will often be the same as the name of the storage provider upon which it is based.
Depending on the storage provider (see below), custom storage pools can be created. In the case of AWS, the ‘ebs’ storage provider supports several configuration attributes:
- ‘volume-type’: volume type (i.e. magnetic, ssd, or provisioned-iops)
- ‘encrypted’: enable/disable disk encryption
- ‘iops’: IOPS per GiB
For example, to provision a 3000 IOPS volume (100GiB x 30IOPS/GiB) by first creating a custom storage pool and then having a newly deployed PostgreSQL use it for its database storage:
juju create-storage-pool iops ebs volume-type=provisioned-iops iops=30 juju deploy postgresql --storage pgdata=iops,100G
See IOPS (Wikipedia) for background information.
Most storage can be dynamically added to, and removed from, a unit. Some types of storage, however, cannot be dynamically managed. For instance, Juju cannot disassociate MAAS disks from their respective MAAS nodes. These types of static storage can only be requested at deployment time and will be removed when the machine is removed from the model.
Certain cloud providers may also impose restrictions when attaching storage. For example, attaching an EBS volume to an EC2 instance requires that they both reside within the same availability zone. If this is not the case, Juju will return an error.
When deploying an application or unit that requires storage, using machine placement (i.e.
--to) requires that the assigned storage be dynamic. Juju will return an error if you try to deploy a unit to an existing machine, while also attempting to allocate static storage.
Assuming the storage provider supports it, storage can be created and attached to a unit using
juju add-storage. Juju will ensure the storage is allowed to attach to the unit’s machine.
Charmed operators can specify a maximum number of storage instances. In the case of the charmed operator ‘postgresql’, a maximum of one is allowed for ‘pgdata’. If an attempt is made to exceed it, Juju will return an error.
Dynamic storage can be detached from units using
Charmed operators can also define a minimum number of storage instances. The postgresql charmed operator specifies a minimum of zero for ‘pgdata’ whereas another charmed operator may specify a different number. In any case, if detaching storage from a unit would bring the total number of storage instances below the minimum, Juju will return an error.
It is not possible to add new storage to a model without also attaching it to a unit. However, with the
juju import-filesystem command, you can add storage to a model that has been previously released from a removed model.
To create a 32GiB EBS volume and attach it to unit ‘ceph-osd/0’ as its OSD storage:
juju add-storage ceph-osd/0 osd-devices=ebs,32G,1
Above, the volume was created in the same availability zone as the instance (a requirement).
To detach OSD device ‘osd-devices/2’ from a Ceph unit:
juju detach-storage osd-devices/2
Detaching storage from a unit does not destroy the storage.
As we saw, detaching storage does not destroy the storage. In addition, when a unit is removed from a model, and the unit has dynamic storage attached, the storage will be detached and left intact. This allows detached storage to be re-attached to an existing unit using
juju attach-storage, or to a new unit using the
--attach-storage flag of
juju deploy or
The underlying cloud’s storage resource is normally destroyed by first detaching it and then using
juju remove-storage. To remove storage from the model without destroying it the
--no-destroy option must be used. Be wary of using the latter option as Juju will lose sight of the volume; it will only be visible from the cloud provider.
If an attempt is made to either attach or remove storage that is currently in use (i.e. it is attached to a unit) Juju will return an error. To remove currently attached storage from the model the
--force option must be used.
Finally, a model cannot be destroyed while storage volumes remain without passing a special option (
--release-storage to detach all volumes and
--destroy-storage to remove all volumes). Naturally, this applies to the removal of a controller as well.
To attach existing storage ‘osd-devices/7’ to existing unit ‘ceph-osd/1’:
juju attach-storage ceph-osd/1 osd-devices/7
To deploy PostgreSQL with (detached) existing storage ‘pgdata/0’:
juju deploy postgresql --attach-storage pgdata/0
-n flags cannot be used together.
To add a new Ceph OSD unit with (detached) existing storage ‘osd-devices/2’:
juju add-unit ceph-osd --attach-storage osd-devices/2
To remove already detached storage ‘osd-devices/3’ from the model. It will also be automatically destroyed on the cloud provider:
juju remove-storage osd-devices/3
To remove currently attached storage ‘pgdata/1’ from the model and prevent it from being destroyed on the cloud provider:
juju remove-storage --force --no-destroy pgdata/1
To upgrade the OSD journal of Ceph unit ‘ceph-osd/0’ from magnetic to solid state (SSD) and dispose of the unneeded original journal ‘osd-journals/0’:
juju add-storage ceph-osd/0 osd-journals=ebs-ssd,8G,1 juju detach-storage osd-journals/0 juju remove-storage osd-journals/0
To destroy a controller (and its models) along with all existing storage volumes:
juju destroy-controller lxd-controller --destroy-all-models --destroy-storage
To destroy a model while keeping intact all existing storage volumes:
juju destroy-model default --release-storage
Assuming the above model was LXD-based, to create a new model and import the released storage volume into it, giving it a storage name of ‘pgdata’:
juju add-model default juju import-filesystem lxd juju:juju-7a544c-filesystem-0 pgdata
The determination of the provider ID (
juju:juju-7a544c-filesystem-0) is dependent upon cloud type. Above, it is given by the backing LXD pool and the volume name (obtained with
lxc storage volume list <lxd-pool>), all separated by a
:. A provider ID from another cloud may look entirely different. The LXD storage provider and associated LXD pools are described in detail below.
Storage management is currently restricted to a single model, which means it is not possible to reuse storage from one model/controller in another. Also, when a model/controller is removed, all associated storage will be destroyed. Support for releasing storage from a model, and enlisting it into another, is planned for a future release.
When upgrading a charmed operator with the juju upgrade-charm command, the existing storage constraints specified at deployment time will be preserved.
It is also possible to change the storage constraints and define new ones by passing the
--storage flag to
juju upgrade-charm. For example, if the ‘pgdata’ storage option did not exist in revision 1 of the postgresql charm, but was introduced in revision 2, when upgrading (from 1 to 2) you could do:
juju upgrade-charm postgresql --storage pgdata=10G
If such a constraint was not provided, ‘rootfs’ would be used (as described in the section on deploying with storage constraints).
Specifying new constraints may be necessary when upgrading to a revision of a charmed operator that introduces new, required, storage options.
There are several cloud-independent storage providers, which are available to all types of models:
- Block-type, creates a file on the unit’s root filesystem, associates a loop device with it. The loop device is provided to the charm.
- Filesystem-type, creates a sub-directory on the unit’s root filesystem for the unit/charmed operator to use. Works with Kubernetes models.
- Filesystem-type, creates a temporary file storage facility that appears as a mounted file system but is stored in volatile memory. Works with Kubernetes models.
Loop devices require extra configuration to be used within LXD. For that, please refer to Loop devices and LXD (below).
AWS-based models have access to the ‘ebs’ storage provider, which supports the following pool attributes:
Specifies the EBS volume type to create. You can use either the EBS volume type names, or synonyms defined by Juju (in parentheses):
- standard (magnetic)
- gp2 (ssd)
- io1 (provisioned-iops)
- st1 (optimized-hdd)
- sc1 (cold-storage)
Juju’s default pool (also called ‘ebs’) uses gp2/ssd as its own default.
- The number of IOPS for io1, io2 and gp3 volume types. There are restrictions on minimum and maximum IOPS, as a ratio of the size of volumes. See Provisioned IOPS (SSD) Volumes for more information.
- Boolean (true|false); indicates whether created volumes are encrypted.
- The KMS Key ARN used to encrypt the disk. Requires encrypted: true to function.
- The number of megabyte/s throughput a GP3 volume is provisioned for. Values are passed in the form 1000M or 1G etc.
For detailed information regarding EBS volume types, see the AWS EBS documentation.
OpenStack-based models have access to the ‘cinder’ storage provider.
The ‘cinder’ storage provider has a ‘volume-type’ configuration option whose value is the name of any volume type registered with Cinder.
MAAS has support for discovering information about machine disks, and an API for acquiring nodes with specified disk parameters. Juju’s MAAS provider has an integrated ‘maas’ storage provider. This storage provider is static-only; it is only possible to deploy charmed operators using ‘maas’ storage to a new machine in MAAS, and not to an existing machine, as described in the section on dynamic storage.
The MAAS provider currently has a single configuration attribute:
- A comma-separated list of tags to match on the disks in MAAS. For example, you might tag some disks as ‘fast’; you can then create a storage pool in Juju that will draw from the disks with those tags.
Azure-based models have access to the ‘azure’ storage provider.
The ‘azure’ storage provider has an ‘account-type’ configuration option that accepts one of two values: ‘Standard_LRS’ and ‘Premium_LRS’. These are, respectively, associated with defined Juju pools ‘azure’ and ‘azure-premium’.
Newly-created models configured in this way use “Azure Managed Disks”. See Azure Managed Disks Overview for information on what this entails (in particular, what the difference is between standard and premium disk types).
Google-based models have access to the ‘gce’ storage provider. The GCE provider does not currently have any specific configuration options.
Oracle-based models have access to the ‘oracle’ storage provider. The Oracle provider currently supports a single pool configuration attribute:
Volume type, a value of ‘default’ or ‘latency’. Use ‘latency’ for low-latency, high IOPS requirements, and ‘default’ otherwise.
For convenience, the Oracle provider registers two predefined pools:
- ‘oracle’ (volume type is ‘default’)
- ‘oracle-latency’ (volume type is ‘latency’).
The regular package archives for Ubuntu 14.04 LTS (Trusty) and Ubuntu 16.04 LTS (Xenial) do not include a version of LXD that has the ‘lxd’ storage provider feature. You will need at least version 2.16. See the Using LXD with Juju page for installation help.
LXD-based models have access to the ‘lxd’ storage provider. The LXD provider has two configuration options:
- This is the LXD storage driver (e.g. zfs, btrfs, lvm, ceph).
- The name to give to the corresponding storage pool in LXD.
Any other parameters will be passed to LXD (e.g. zfs.pool_name). See upstream LXD storage configuration for LXD storage parameters.
Every LXD-based model comes with a minimum of one LXD-specific Juju storage pool called ‘lxd’. If ZFS and/or BTRFS are present when the controller is created then pools ‘lxd-zfs’ and/or ‘lxd-btrfs’ will also be available. The following output to the
juju storage-pools command shows all three Juju LXD-specific pools:
Name Provider Attributes loop loop lxd lxd lxd-btrfs lxd driver=btrfs lxd-pool=juju-btrfs lxd-zfs lxd driver=zfs lxd-pool=juju-zfs zfs.pool_name=juju-lxd rootfs rootfs tmpfs tmpfs
As can be inferred from the above output, for each Juju storage pool based on the ‘lxd’ storage provider there is a LXD storage pool that gets created. It is these LXD pools that will house the actual volumes.
The LXD pool corresponding to the Juju ‘lxd’ pool doesn’t get created until the latter is used for the first time (typically via the
juju deploy command). It is called simply ‘juju’.
lxc storage list is used to list LXD storage pools. A full “contingent” of LXD non-custom storage pools would like like this:
+------------+-------------+--------+------------------------------------+---------+ | NAME | DESCRIPTION | DRIVER | SOURCE | USED BY | +------------+-------------+--------+------------------------------------+---------+ | default | | dir | /var/lib/lxd/storage-pools/default | 1 | +------------+-------------+--------+------------------------------------+---------+ | juju | | dir | /var/lib/lxd/storage-pools/juju | 0 | +------------+-------------+--------+------------------------------------+---------+ | juju-btrfs | | btrfs | /var/lib/lxd/disks/juju-btrfs.img | 0 | +------------+-------------+--------+------------------------------------+---------+ | juju-zfs | | zfs | /var/lib/lxd/disks/juju-zfs.img | 0 | +------------+-------------+--------+------------------------------------+---------+
The three Juju-related pools above are for storing volumes that Juju applications can use. The fourth ‘default’ pool is the standard LXD storage pool where the actual containers (operating systems) live.
To deploy an application, refer to the pool as usual. Here we deploy PostgreSQL using the ‘lxd’ Juju storage pool, which, in turn, uses the ‘juju’ LXD storage pool:
juju deploy postgresql --storage pgdata=lxd,8G
See Using LXD with Juju for how to use LXD in conjunction with Juju, including the use of ZFS as an alternative filesystem.
Kubernetes-based models have access to the ‘kubernetes’ storage provider, which supports the following pool attributes:
The storage class for the Kubernetes cluster to use:
The Kubernetes storage provisioner. For example:
Extra parameters. For example:
Using storage with Kubernetes is covered on the Persistent storage and Kubernetes page.
LXD (localhost) does not officially support attaching loopback devices for storage out of the box. However, with some configuration you can make this work.
Each container uses the ‘default’ LXD profile, but also uses a model-specific profile with the name
juju-<model-name>. Editing a profile will affect all of the containers using it, so you can add loop devices to all LXD containers by editing the ‘default’ profile, or you can scope it to a model.
To add loop devices to your container, add entries to the ‘default’, or model-specific, profile, with
lxc profile edit <profile>:
... devices: loop-control: major: "10" minor: "237" path: /dev/loop-control type: unix-char loop0: major: "7" minor: "0" path: /dev/loop0 type: unix-block loop1: major: "7" minor: "1" path: /dev/loop1 type: unix-block ... loop9: major: "7" minor: "9" path: /dev/loop9 type: unix-block
Doing so will expose the loop devices so the container can acquire them via the
losetup command. However, it is not sufficient to enable the container to mount filesystems onto the loop devices. One way to achieve that is to make the container “privileged” by adding:
config: security.privileged: "true"
For guidance on how to create a charmed operator that uses these storage features see Writing charmed operators that use storage.
Last updated 18 days ago.